import type { languages } from 'monaco-editor';

export function buildMonarchLanguage(): languages.IMonarchLanguage {
  return {
    defaultToken: '',
    tokenPostfix: '.kql',

    brackets: [
      { open: '{', close: '}', token: 'delimiter.curly' },
      { open: '(', close: ')', token: 'delimiter.parenthesis' },
    ],

    tokenizer: {
      root: [
        // Whitespace
        [/[ \t\r\n\u00A0]+/, 'white'],

        // Quoted strings
        [/"/, { token: 'string.quote', next: '@string' }],

        // Operators (longer patterns first)
        [/<=/, 'operator'],
        [/>=/, 'operator'],
        [/[<>:]/, 'operator'],

        // Brackets
        [/[{}]/, 'delimiter.curly'],
        [/[()]/, 'delimiter.parenthesis'],

        // Literals
        [/\b(?:null|true|false)\b/, 'keyword.literal'],

        // Conjunction keywords (all case combinations — Monarch ignores rule-level /i):
        // and/or must be followed by whitespace, not too.
        [/[aA][nN][dD](?=[ \t\r\n\u00A0]|$)/, 'keyword'],
        [/[oO][rR](?=[ \t\r\n\u00A0]|$)/, 'keyword'],
        [/[nN][oO][tT](?=[ \t\r\n\u00A0]|$)/, 'keyword'],

        // Numbers
        [/\d+(?:\.\d+)?/, 'number'],

        // Escaped characters (e.g. \: \( \and \uXXXX)
        [/\\./, 'identifier'],

        // Wildcard
        [/\*/, 'operator.wildcard'],

        // Unquoted literal run (field names like machine.os, bare values)
        [/[^\s\\():<>"*{}]+/, 'identifier'],
      ],

      string: [
        [/[^\\"]+/, 'string'],
        [/\\(?:u[0-9a-fA-F]{4}|[\s\S])/, 'string.escape'],
        [/"/, { token: 'string.quote', next: '@pop' }],
      ],
    },
  } as languages.IMonarchLanguage;
}